Good logging for me involves: - aggregation from multiple systems - one-line per event (so I can use unix tools) - (ideally) enough information to replay som

Good logging for me involves: - aggregation from multiple systems - one-line per event (so I can use unix tools) - (ideally) enough information to replay some action in the event of a bug - being able to live-tail it within seconds - putting a unique, grepable identifier next to each printed output